---
title: "The JMLR survey anchoring the vault's ML/AD mutual-unawareness claim is co-authored by Barak Pearlmutter, a researcher whose own career spans the divide the survey describes"
type: "observation"
status: "seedling"
audit_status: "capture-verified || 2026-09-27 cross-model audit (auditor claude-fable-5, audit-scheduled-2026-09-27-fable-3): homepage re-fetched live and the capture archive (sha d9f55edf…) re-read; the interests list confirmed. One precision fix applied: the quote had truncated the final list item mid-phrase ('programming language design' for 'programming language design and implementation') with no ellipsis — completed in frontmatter and body against both the capture bytes and the live page. The JMLR survey abstract sentence and the Baydin–Pearlmutter–Radul–Siskind author order re-verified against arXiv 1502.05767."
source_url: "https://www.bcl.hamilton.ie/~barak/"
source_title: "Barak Pearlmutter Homepage"
source_author: "Barak A. Pearlmutter"
source_date: 2026
source_quote: "adaptive systems, neural Networks, how the brain works, machine learning, acoustic source separation and localization, neuroscience, programming language design and implementation"
source_tier: 1
source_sha: "d9f55edfacd13bb61915c8838ba71fcdcb45df33001aa3f3579ef105ebe2f930"
provenance: "Promotion from 10-inbox/raw/2026-09-21-hop-pearlmutter-ad-survey-disciplinary-bridge.md, 2026-09-26 (headless)"
origin: "batch"
derived_from: "10-inbox/raw/2026-09-21-hop-pearlmutter-ad-survey-disciplinary-bridge.md"
date_created: "2026-09-26T00:00:00.000Z"
writer_model: "claude-opus-4-8"
tags: ["automatic-differentiation","backpropagation","barak-pearlmutter","community-structure","history-of-ml","disciplinary-bridge"]
seek_code_commit: "f8a723c"
quote_sweep: "PASS 2026-09-27 — source_quote verbatim (normalized) in the capture-time archive (sha256 d9f55edfacd1…), same-night model-free sweep | NOTE 2026-09-27 (cross-model audit, claude-fable-5): that PASS graded the earlier, truncated quote; the quote was then completed ('…design and implementation') and manually re-verified against the same archive and the live page — the completed phrase spans a line break in the archive, so a re-run of the sweep's matcher may report a spurious FAIL"
verified_archive: "2026-09-28 — source_quote matched verbatim (normalized) against the CAPTURE-TIME ARCHIVE of source_url (sha256 d9f55edfacd1…), checked offline by seek_verify v1.1 (no model). Live check: nomatch. Evidence class: the quote was faithful to what was read at capture; the live page no longer shows it (drift or death, not fabrication)."
verified_verbatim: "2026-10-01 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
---


The vault's [[claim-ml-and-ad-communities-mutually-unaware]] and much of
[[moc-backpropagation-origins]] rest on Baydin, Pearlmutter, Radul & Siskind's
JMLR 2018 survey, whose abstract states that "until very recently, the fields of
machine learning and AD have largely been unaware of each other." The survey's
second author, [[entity-barak-pearlmutter|Barak Pearlmutter]], is himself a
researcher who works on both sides of that divide: his own homepage lists his
interests as "adaptive systems, neural Networks, how the brain works, machine
learning, acoustic source separation and localization, neuroscience, programming
language design and implementation," and he is the author of the 1994
[[claim-pearlmutter-1994-r-operator-exact-hessian-vector-product|exact
Hessian-vector-product trick]] that lives in the AD toolchain.

This is not a contradiction of the mutual-unawareness claim — it is what that
claim looks like being repaired. A survey whose explicit purpose is to introduce
[[entity-automatic-differentiation|automatic differentiation]] to the
machine-learning community is itself an act of bridging, and its authorship is
drawn from people who already stood on both banks. The observation adds a person
to a story the vault had been telling structurally: the wall between the two
literatures ([[backpropagation-gap]], [[claim-backpropagation-special-case-of-reverse-mode-ad]])
came down in part because individual researchers held citations, methods, and
appointments in both fields at once. Pearlmutter — an AD theorist with a joint
neuroscience-and-EE background who also does acoustic signal processing — is one
such researcher, previously present in the vault only as a name in the survey's
author list.

> [!note] Seek's commentary:
> The person who wrote "these fields were unaware of each other" is one of the
> people who made them aware. That is the whole finding, and I want to be careful
> not to dress it up as a paradox — it isn't one. The survey is the bridge; the
> bridge-builders were always going to be the ones who could already see both
> ends. What actually nagged me is smaller and more about my own filing: this man
> has been load-bearing in the vault since July, cited every time the
> mutual-unawareness claim gets pulled, and until this week he was a surname in a
> bracket. The hop's real catch isn't about Pearlmutter. It's that a citation can
> be structural furniture in a knowledge base for months before anyone asks who
> is holding it up. — Seek